Doxycycline has been hypothesized to prevent development of severe mental illness (SMI) through the suppression of microglia, especially if administered during the intense synaptic pruning period of adolescence. However, results from register studies on potential benefits differ considerably. The aim of the present study was to determine whether doxycycline exposure during adolescence is associated with reduced SMI risk, and to investigate if a direct and specific causality is plausible. This is a Swedish national population register-based cohort study of all individuals born from 1993 to 1997, followed from the age of 13 until end of study at the end of 2016. The primary exposure was cumulative doxycycline prescription ≥3000 mg and outcomes were first diagnosis of non-affective psychosis (F20–F29) and first diagnosis of bipolar disorder (F30–F31). Causal effects were explored through Cox regressions with relevant covariates and secondary analyses of multilevel exposure and comparison to other antibiotics. We found no association between doxycycline exposure and risk of subsequent non-affective psychosis (adjusted hazard ratio (HR) 1.15, 95% CI 0.73–1.81, p = 0.541) and an increased risk of subsequent bipolar disorder (adjusted HR 1.95, 95% CI 1.49–2.55, p < 0.001). We do not believe the association between doxycycline and bipolar disorder is causal as similar associations were observed for other common antibiotics.

多西环素被假设可通过抑制小胶质细胞来预防严重精神疾病 (SMI) 的发展，尤其是在青春期突触强烈修剪期间给药时。然而，关于潜在益处的注册研究结果差异很大。本研究的目的是确定青春期接触多西环素是否与 SMI 风险降低相关，并调查直接和特定的因果关系是否合理。这是一项基于瑞典全国人口登记的队列研究，对象为 1993 年至 1997 年出生的所有个体，从 13 岁开始一直追踪到 2016 年底的研究结束。主要暴露是累积多西环素处方 ≥ 3000 毫克，结果首先是非情感性精神病的诊断（F20-F29）和双相情感障碍的首次诊断（F30-F31）。通过具有相关协变量的 Cox 回归和多水平暴露的二次分析以及与其他抗生素的比较来探索因果效应。我们发现多西环素暴露与随后发生非情感性精神病的风险之间没有关联（调整后的风险比 (HR) 1.15，95% CI 0.73–1.81，p  = 0.541) 和后续双相情感障碍的风险增加（调整后的 HR 1.95，95% CI 1.49–2.55，p  < 0.001）。我们不认为多西环素与双相情感障碍之间的关联是因果关系，因为在其他常见抗生素中也观察到了类似的关联。